• Nie Znaleziono Wyników

Turning the light on in Virginia: New perspectives on choice behavior modeling

N/A
N/A
Protected

Academic year: 2021

Share "Turning the light on in Virginia: New perspectives on choice behavior modeling"

Copied!
26
0
0

Pełen tekst

(1)

Prof.dr.ir. C.G. Cliorus

Turning the light on in

Virginia: New perspectives

on choice behavior modeling

Intreerede 5 november 2014

4

(2)

Turning the light on in Virginia:

New perspectives on choice

behavior modeling

Intreerede

Uitgesproken op 5 november 2014 ter gelegenheid van de aanvaarding van

het ambt van hoogleraar ''Choice behavior modeling'' aan de Faculteit Techniek, Bestuur en Management van de Technische Universiteit Delft.

door

(3)
(4)

Mijnheer de Rector Magnificus, leden van het College van Bestuur,

Collegae hoogleraren en andere leden van de universitaire gemeenschap, Zeer gewaardeerde toehoorders,

Dames en heren.

A series of choices^

Life is a series of choices, great and small. And even larger than life concepts such as international trade balances, flue pandemics, and...traffic jams, can be seen as the aggregate result of choices made by ordinary people like you and me. Hence it should come as no surprise that understanding and predicting human choice behavior has always been one of the most important goals of the Social Sciences. And when someone who is educated in a place like Delft says 'understanding and predicting', he or she thinks 'modeling', and preferably using mathematics. Luckily - for Delftenaren at least - in the late 1970s a new research field was born^, that concerns itself with the development of mathematical choice models. Some ofthe baby's crucial first steps were actually made - and not coincidentally so - at another university of technology: MIT, in Cambridge, Massachusetts (Ben-Akiva, 1973). This makes perfect sense given the notion that over the past few decades, many prominent universities of technology, including TU Delft, have come to understand that they could, and should, make important contributions to the Social Sciences as well, using their highly effective toolbox of mathematical modeling skills.

This lecture has some goals: first, I want to convince you of the importance and beauty of my research field; and I want to show you what an interesting future lies ahead for choice modelers. Third, I want to explain why Delft, and in particular its Faculty of Technology, Policy and Management, is the perfect place for hosting a team of choice modelers. Fourth, if you are a scholar or a student, I would like to entice you to join forces with me and my colleagues; and if you are family or friend, my goal is also to provide a bit of entertainment - to the extent possible, that is.

As I alluded to, people make hundreds of choices each day. Some are big and well contemplated, while others are small and made in a routine fashion. However, what all these different choices have in common, is that they can be used to infer underlying preference trade-offs. For example, the travel choices you made when coming to this Auditorium, echo your preference for certain travel modes over others; your aversion (or lack thereof) for being late; your ^ / would like to thank Prof. Michel van Eeten for making a range of insightful

comments concerning a previous version of this lecture.

(5)

dislike of having to search for a parking spot, etc.; and your choices also echo how you trade off these often diverging preferences. The core research aim of choice modelers is to infer these preferences and trade-offs from your observed choice behavior, and subsequently, to predict your future choices based on these inferences. To do this, choice modelers use mathematical and statistical models, and of course large amounts of data to estimate and validate them.

One ofthe beauties of my research field is its immense societal relevance (e.g. Nobelprize.org, 2000): choice models are routinely used in a wide variety of practical settings, to predict traffic jams and election outcomes; to determine the optimal price for game consoles or medical treatments; to determine optimal investment levels for nature reserves and waterworks, you name it (Hess & Daly, 2014). Choice models can be considered the micro-level foundations beneath many macro-level policies, in sectors ranging from transportation, environment, marketing, to health, and beyond.

A paradox

Another aspect of beauty is that the entire field is based on a paradox: how on earth can anyone even aspire to mathematically model human

(6)

choice behavior? Isn't our behavior far too complicated to be captured in a mathematical model? Although for some of your behaviors the answer to this question is probably 'yes', in my view the paradox between human behavior and mathematical models is actually nothing but a paradox: a seeming contradiction.

PARADOX

1'. Ym + exp[/?,n • (^fcm (Pi - 1) • •1 ~Xim)] PmXir

Behavioral scientists, more than anything, are interested in understanding human behavior; their central research question being 'how and why do people make the choices they make?'. They routinely criticize their choice modeling colleagues, whose main research goal is to predict what choices people make, and who presumably are far too much interested in mathematical elegance, leading to models that wrongfully depict humans as computers. Choice modelers, in turn, criticize behavioral scientists for failing to come up with tractable, usable mathematical models of human behavior. They laugh at them, and remark that the ever growing collection of subtle behavioral phenomena that behavioral scientists bring to the fore, reminds one of a collection of used stamps: nice to look at, but you can't use them to mail letters.

Although mutual criticism is of course a necessary ingredient for good science, my impression is that the lack of understanding and appreciation between

(7)

behavioral scientists and choice modelers damages, and perhaps even jeopardizes, the future of my research field.

The current situation in a way makes me think of the political climate in the United States, where competition between the two main parties has reached a point where arguments hardly play a role anymore, and where collaboration is often seen as betrayal. Given this correspondence, who better to consult for a solution to this stalemate, than a US President who has lived - and thrived - in a decade of fervent political and culture wars?

Turning the light on in Virginia

When Bill Clinton arrived in Alexandria, Virginia, the local Democratic Party was facing the prospect of a cruel '94 mid-term election. It was under ongoing attacks from their right-wing adversaries, and as they responded with attack ads themselves, Virginia politics gradually turned into a very dark place. At the end of his thirty minute speech, Clinton offered a way out of this warzone, as he proclaimed (Clinton, 1994): "What you have to do [...] is not so much to bash your adversaries, although, goodness knows, you need to answer them back. You need to turn the light on in Virginia and let the light shine and let people feel the future flowing through their veins [...]'' I don't need to tell you how the room reacted.

(8)

So what message has this political giant in store for us, choice modelers? He tells us that waging the occasional battle with behavioral scientists is OK, but that it is much more important to present the research community an inspiring perspective on the future. Now what kind of future should that be? Clinton, as a president, chose to take some key elements from his opponents' program, such as a conservative approach to manage the budget; and to combine these - e t e n w i t s ^ h his cmn party's ideasr^lthough it took some time to take o.., this integrative approach made his second term as a President one of the most prosperous in US history.

Back to choice modeling: my view of the field's future is that we, choice modelers, should try and inject more human behavior into the veins of our models, rather than simply laugh at behavioral scientists and their collection of used stamps. This solution of course, is kind of obvious, and you really don't need a professor to bring it to your attention. And note that I am certainly not the first who is making this point. But I do dare to say that my group at TU Delft is a visible part of a still relatively small - but growing and increasingly influential - circle of choice modelers, who devote their careers to doing just that: improving choice models by increasing their behavioral realism, without sacrificing mathematical elegance. This is why, rather than being a professor of choice modeling, I chose to become professor of choice behavior modeling, with an emphasis on choice, on behavior, and on modeling.

Time to get down to business: what am I talking about, when I refer to 'Increasing the behavioral realism of choice models"? Let me start by briefly going into an example of past work, before I move on to discuss much needed future research.

Regretting utility

An important example of how a small group of choice modelers have succeeded in increasing the behavioral realism of their models, relates to how we model the translation of preferences into choices. The heuristic or choice rule that is assumed in the overwhelming majority of choice models is that of so-called linear-additive utility maximization (e.g., Ben-Akiva & Lerman, 1985; Train, 2009). This rule postulates that people choose the maximum utility alternative from the choice set, which obviously makes a lot of sense and can even be considered a circular reasoning.

Another postulate that is embedded in this choice rule, however, is much more tricky: the utility associated with an alternative, say a car, is postulated to

(9)

depend only on its own performance, not on that of competing alternatives. This notion contrasts with an abundance of evidence, mostly arising from the field of consumer psychology, that actual choice behavior is fundamentally at odds with this postulate (e.g., Simonson, 1989; Tversky & Simonson, 1993). In fact, when faced with a choice set - of, for example, breakfast cereals, or car types - consumers are routinely found to evaluate a considered alternative using the performance of other alternatives as reference points or bench-marks. One consequence of this so-called reference dependency is that choice behavior can be influenced by subtle changes in the choice set. At an abstract level: the relative popularity of product A compared to B can be altered by cleverly positioning another product C in the choice set. Marketeers were of course keen to capitalize on this insight. I can guarantee you that the last time you bought a cereal package, or a car, your choice has been influenced by these so-called choice set engineering techniques.

How did most choice modelers react to this large and growing body of behavioral findings? You may guess the answer, by now. They said something like: ''well, that is all very interesting, but a) your results are in conflict with the core axioms of choice models; b) even if they were not, there is no way to capture this kind of awkward behavior into tractable mathematical models; c) enjoy the rest of

(10)

Choice set engineering

your day". Luckily, a small group of choice modelers from the marketing field agreed with a) and with c), but not with b). Like my daughters' idols K3, they said something like ''Gaat niet; Bestaat niet!" Starting in the late 1980s, they started to develop alternative choice models that did incorporate this awkward reference-dependency into mathematical models. In my view, they succeeded partly: my own experience is that most alternative models (e.g., Kivetz et al., 2004) are too complicated mathematically, and come with some issues that make them difficult to use in practice, especially by people without very strong econometric skills. Nonetheless, these models did certainly help inspire a second generation of choice models that explicitly focused on mathematical tractability. One of these models is the random regret minimization model which I developed over the past few years, starting at TU Eindhoven (Chorus et al., 2008), continuing here in Delft (Chorus, 2010; 2014), and with help of more than a handful of scholars who are in this room today. This model has been explicitly constructed to allow non-hardcore econometricians, like me, to use it successfully. The random regret model assumes that people, when choosing, try to avoid regret. They do so by comparing each option which all other options, on every criterion, such as cost or quality; in other words, the random regret model postulates that the performance of non-chosen options influences the

(11)

evaluation of a chosen option. As a consequence, the model's mathematical form captures reference-dependency ofthe type described above, and it allows for the analysis of choices from engineered choice sets. But at the same time, the model consumes no more parameters than the classical random utility model; neither is it fundamentally more complicated than conventional models. It is my impression that the mathematical tractability of the random regret

-TTriTTimTzation model has been one important reason why it has now been incorporated in some of the most widely used econometric software packages worldwide (e.g.. Econometric Software, 2012; Statistical Innovations, 2014).

Maximum Utility Model

Minimum Regret Model

I Laten I

statistics

Sawtooth Software

Although many of you may expect - or fear - that I will now embark on a lengthy discussion of this personal hobby-horse of mine, I will not do so. Clearly, the random regret model is one ofthe reasons why I am standing here today; and my group will remain actively involved in regret modeling for some years to come (e.g. van Cranenburgh et al., in press). However, I would like to use the remainder of my talk to argue what I think needs to be done the coming years, rather than looking back at past and current research. But let it be noticed that in my view the random regret model, as well as other models that have more recently been proposed (e.g. Leong & Hensher, 2014), certainly help prove those scholars wrong, who claimed that awkward choice behavior necessarily leads to awkward choice models.

(12)

Altruism, fairness, social responsibility

Now, back to the future. Looking at the coming five to ten years, I see several important directions for future research. A first direction concerns a further injection of behavioral realism into choice models, in the form of taking notions such as altruism, fairness, and social responsibility into account. Conventional choice models are based on the idea that people are selfish, and that their preferences - and hence their choices - are purely seir-regardlng. However, many social scientists (including some leading economists) have convincingly argued for many years, that people care about others and that this behavioral phenomenon needs to be reflected in economic models (e.g., Fehr & Schmidt, 1999; Fehr & Fischbacher, 2003). Think of someone who gives up his job to follow his partner who pursues a career overseas. Or think of someone who chooses not to join a protest march against the arrival of Volkert van der G. in his town, simply because he believes that Volkert too needs a place to live. Or think of someone who buys an expensive electric car, hoping to contribute to a more sustainable environment.

In fact, I believe that some of the most important and consequential choices people make, are largely driven by notions such as altruism, fairness, and social responsibility. Hence, if we ignore these factors in our choice models, this leads

(13)

to a seriously biased understanding of how people behave. Once again, you don't need to be a professor, or even hold a PhD, to come to this rather obvious conclusion. Then why is this truism not being captured in our choice models? An important reason is that it makes the life of a modeler so much easier. Take the commensurability axiom, which suggests that a Euro spent for purpose A weighs as much as a Euro spent on purpose B. This axiom is extremely converrient for the development of economic models. However convenient, the axiom severely conflicts with empirical travel behavior research which finds that for many road users, a Euro spent on road pricing weighs much heavier than a Euro spent on petrol (e.g., Hensher et al., 2007); the reason presumably being that many people find the concept of road pricing simply unfair (e.g., Jakobsson et al., 2000). On the other hand, many citizens support road pricing for reasons of environmental concern (e.g., Bliemer et al., 2008), even if economic theory suggests they should not, as implementation would be quite disadvantageous for their personal travels.

As another example of how other-regarding preferences may co-determine choice behavior, a recent study (Manville & Cummins, in press) finds that most Americans who support investments in Public Transport do not do so for their own good - as you know, many Americans would rather die than board a train; their support is mostly based on broad societal concerns. When such societal

€ > €

On the one hand...

on the other

QROEN

(14)

motivations are ignored in the process of developing and evaluating transport policies - and a 'selfish consumers-perspective is adopted instead - this is likely to have a large and negative impact on investments in Public Transportation.

1

K L A l J n S ! i J Poor \ \ ^ ^ ^ ^ ^ i S o v e r e i g n % 1; C i t i z e n I

mm

^

^burgerschap

samen stad zijn

1

^ ^ ^ ^ ^ i S o v e r e i g n % 1; C i t i z e n I

mm

^

^burgerschap

samen stad zijn

For a variety of methodological reasons, incorporating these subtle and sometimes paradoxical phenomena in choice models is a difficult task. Yet, sweeping them under the rug of convenience of course does not make them go away; instead, it would mainly lead to counter effective policies. And perhaps, counter effective is not even a strong enough term, as it focuses mainly on the accuracy of models and policies while the actual problem runs deeper than that. I believe that policies which are based on the notion that our choices are only based on self-regarding preferences - this is called the principle of consumer-sovereignty - are flawed in a deep, moral sense of the word. Ultimately, such a worldview may even end up being a self-fulfilling prophecy as it may instill a belief among citizens that they are, above all, consumers^ I see an important

^ See Kelman (1981) for a profound discussion of die severe moral consequences that follow from framing individuals merely as consumers (as opposed to citizens) in Cost Benefit Analyses. In the Netherlands, as in many other countries. Cost Benefit Analyses are routinely used to evaluate major infrastructure investments and policies; they are more often than not based on the outcomes of choice models. Hence, if one wants these investments and policy decisions to be based not merely on a 'consumer'perspective, but rather on a broader 'citizen'perspective, a crucial step is to incorporate notions of 'citizen- sovereignty'in the choice models which underlie Cost Benefit Analyses.

(15)

responsibility for choice modelers here, as it is our research which may enable the methodological and empirical advances which are needed to accommodate other-regarding preferences in our models and policies. In other words, choice modelers have a crucial role in enabling a change from consumer-sovereignty to what I would like to call 'citizen-sovereignty'^

1\lüLe ÜidL several colleagues at the TLO group are working on similar themes, from the perspective of Cost Benefit Analysis and transport policy (e.g., van Wee, 2011). When our choice modelers join forces and also involve governance experts within our Faculty, I believe that we can make a lasting contribution to this line of research.

Choice models and BigData

Another, very enticing, future of choice behavior modeling lies in teaming up with fellow researchers who are working with so-called BigData. If there is one defining aspect of how Social Science today differs from Social Science yesterday, it is the availability of enormous amounts of data. Combine this with steeply increasing computing power, and the situation emerges where behavioral patterns of millions of people can be monitored and predicted time (e.g. Giannotti et al., 2013). Think of a flu pandemic which is tracked real-time by means of monitoring where and when people Google-search 'aspirin' (Cook et al., 2011). Or think of the increase in predictive power of mobility and traffic models, caused by tracking mobile phone signals of road users (e.g., Candia et al., 2008; Steenbruggen et al., 2013). Although many scholars, also here at TU Delft, argue that everything starts and ends with data (the Dutch refer to this as "meten is weten"), I tend to prefer Biglnsights over BigData; it is my strong conviction that without good theories and models of behavior, tapping BigData may even work counter effective. Choice models, being firmly rooted in behavioral theories, may be just the ingredient that is needed to turn Data into Insights.

What makes the endeavor of combining choice models and BigData particularly challenging, is that choice models as we know them need very clean and detailed data, whereas most sets of BigData are usually quite 'messy', and in crucial ways not detailed enough. This mismatch triggers a fundamental rethink of choice models and the data they consume. I consider this one of the most important directions for further research in our field.

^ See Nyborg (2000) for the formal derivation of indifference curves for individuals whose preference-structures differ, depending on whether they wish to behave as consumers, or citizens. Her approach can form a valuable building block for the derivation of the type of choice models I refer to.

(16)

As alluded to, this combination of choice models and BigData holds the promise of a much deeper understanding and hence a much better prediction of choice behavior. However, for many policy-makers and marketing managers, the Holy Grail is to not so much to understand and predict, but to influence behavior. They too believe that the advent of BigData and its clever combination with mathematical models of behavior provide new and exciting opportunities for them (e.g.. Brown et al., 2011). Thihk of a personalized travel information service that tracks a traveler's every movement using GPS, and has access to her Outlook agenda; the service then uses choice models to infer the traveler's preferences from her observed choice behavior (e.g., Arentze, 2013), and merges these insights with her schedule for the day; the result is a tailor made advice (e.g. Chorus et al., 2009). Now picture this situation for not only one, but ten million travelers. If owned by a government, such a service would be able to optimize traffic flows like never before, as it may advise travelers to behave in ways that are sometimes detrimental to them personally, but are optimal for the system as a whole (e.g., Wu & Huang, 2010; Avineri, 2012). If owned by a firm in which McDonald's has a stake, one can imagine that more traffic will be sent towards routes that feature one of their restaurants (I wouldn't mind, by the way - but I am sure some of you would).

(17)

Of course, even without choice models, governments and firms have ample opportunities to influence behavior. However, knowing preferences of citizens and consumers clearly gives governments and firms more, and more powerful, opportunities to manipulate. I have spoken with too many policy-makers and marketing managers, to be naive in this regard: if given the opportunity, governments and firms will exploit the powers given to them. This raises many ethical and governance related questions concerning the combination of BigData and choice models, that go way beyond the typical focus on privacy matters. I cannot wait to explore the enticing and at the same time frightening prospect of transforming BigData into Biglnsights using choice models, jointly with BigData specialists. Transportation engineers, governance experts, and philosophers at the TPM Faculty, TU Delft, and beyond.

The world is your oyster!

To conclude this substantive part of my lecture, I would like to note that while I mainly discussed choice behavior modeling from a research perspective, there is a very obvious and strong teaching element here as well. For years, the TLO group teaches choice modeling at the Bachelor, Master, and PhD-level,

(18)

and various new lectures and even courses are currently in the making. Since TU Delft rightfully wants its education to be rooted in research, our students will be seeing many of the here described research ideas in their courses, in due time. I hope and expect that my lecture has convinced you that choice behavior modeling, besides being a fascinating research field, may also be a very valuable part of today's engineers' training. If you are a TU Delft graduate who not only has learned to build a bridge, but also to engineer a choice set -the world is your oyster!

Dankwoord

The setting of an inaugural lecture seems to be designed in a way as to make the audience think that the professor has done his work all by himself. But, let me be clear, there are many people I must thank for helping make possible that I stand here today. First, I want to thank Prof. Karel Luyben, rector of this university, for having the confidence to promote a young guy like me to a full professorship. Similarly, I want to thank professor Theo Toonen for his faith in me. Professor Toonen runs a fascinating Faculty, and I am proud to be a part of it. Also professor Yao-Hua Tan and his successor professor Paulien Herder deserve thanks, for supporting me at crucial moments. More generally speaking, during my burn-out last year, I experienced that my superiors at TU Delft are not only with me when there is something to celebrate, but also when things are not going so well. Thank you.

(19)

I have the privilege to lead a wonderful group. TLO provides strong evidence that top-notch academic performance can coincide with high doses of inappropriate humour and a warm and supportive environment. Many TLO-colleagues deserve to be thanked here, but for reasons of strict time regulations I focus on one. Bert, professor van Wee, you have been a mentor for me from the moment you started supervising my PhD-research. Most big trees cast large shadows, in which it is difficult to grow, but you give ample room to the next generation, which is an inspiration to me. Thank you.

Both coming from a family of scholars, my parents, Rogier and Mirjam, have at crucial moments convinced me that a career in academics is a worthy pursuit. Your support, and the fact that we are growing closer as time goes by, means a lot to me. I also thank my sister Sarah, my brother Felix, and Quirine, as well as my lovely in-laws, for their warm interest in me - both personally and career-wise; including my citation counts.

It feels really good to wear this more than 65 year old gown which my grandfather, in the 1940s, and my uncle, in the 1980s, wore during their inaugural at Leiden University. As you can see, these pictures get bigger through the years. This is purely a matter of pixel-numbers, and of course does not imply that Chorus professors have grown taller in subsequent generations; on the contrary! Nonetheless, I can almost hear the gown say that he is very, very pleased to be finally moving from Leiden to Delft now, since this implies a significant move upwards in the official Times Higher Education Reputation ranking^!

Tot slot wil ik het woord richten tot Nienke, Cato en Loes. Ik heb vroeger eens in een dronken bui aan een huisgenoot verteld dat ik wilde trouwen met een katholieke onderwijzeres, en vier dochters met haar krijgen. Een buitenstaander die goed kan rekenen zou dus kunnen zeggen, dat deze jongensdroom half is uitgekomen. Maar die weet dan niet wat hij zegt. Nienke was inderdaad tot voor kort onderwijzeres, en ze is nog steeds een beetje katholiek; maar ze is altijd zoveel meer geweest, voor mij. Ik heb het repertoire van de Pet Shop Boys (1987) nodig om in een paar woorden te zeggen wat ik voor je voel: "rm so happy that you're mine". En dan mijn dochters, wat een prachtmeiden zijn jullie. En lachen dat ik met jullie kan. Wanneer ik me weer eens verlies in wetenschappelijke onzinpraat tijdens het avondeten, wat is er dan mooier dan jullie te horen zeggen: "mam, de hoogleraar doet een beetje raar." - en zo is het! Ik heb gezegd.

' See:

http://www. timeshighereducation. co. uk/world-university-rankings/2014/reputation-ranking/range/Ol -50 http.y/www.timeshighereducation.co.uk/world-university-rankings/2014/reputati

(20)
(21)

References

Arentze, T. A. (2013). Adaptive Personalized Travel Information Systems: A Bayesian Method to Learn Users' Personal Preferences in Multimodal Transport Networks. IEEE Transactions on Intelligent Transportation

Systems, 14(4), 1957-1966

Avineri, E. (2012). On the use and potential of behavioural economics from Üie perspective of transport and climate change. Journal of Transport

Geography, 24, 512-521.

Ben-Akiva, M. (1973). The structure of travel demand models. Phd-thesis,

Massachusetts Institute of Technology, Cambridge, MA.

Ben-Akiva, M. E., & Lerman, S. R. (1985). Discrete choice analysis: theory and

application to travel demand. MIT press.

Bliemer, M., Steg, L., & Van Wee, B. (2008). Pricing in road transport: a

multi-disciplinary perspective. Edward Elgar Publishing.

Brown, B., Chui, M., & Manyika, J. (2011). Are you ready forthe era of'big data'.

McKinsey Quarterly, 4, 24-35.

Candia, J., Gonzalez, M. C, Wang, P., Schoenharl, T, Madey, G., & Barabasi, A. L. (2008). Uncovering individual and collective human dynamics from mobile phone records. Journal of Physics A: Mathematical and Theoretical, 41(22), 224015.

Chorus, C. G., Arentze, T. A., & Timmermans, H. 1 (2009). Traveler compliance with advice: a Bayesian utilitarian perspective. Transportation Research

Part E: Logistics and Transportation Review, 45(3), 486-500.

Chorus, C. G., Arentze, T. A., & Timmermans, H. J. (2008). A random regret-minimization model of travel choice. Transportation Research Part B:

Methodological, 42(1), 1-18.

Chorus, C. G. (2010). A new model of random regret minimization. European

Journal of Transport and Infrastructure Research, 10{2), 181-196

Chorus, C.G. (2014). A generalized random regret minimization model.

Transportation Research Part B, 68, 224-238

Clinton, W.J. (1994). Remarks at the Kennedy-King Dinner in Alexandria,

Virginia

October 21, 1994

Cook, S., Conrad, C, Fowlkes, A. L., & Mohebbi, M. H. (2011). Assessing Google flu trends performance in the United States during the 2009 influenza virus A ( H l N l ) pandemic. PloS one, 6(8), e23610.

Econometric Software (2012). NLOGIT Version 5.0 Reference Guide. Plainview, NY

Fehr, E., & Schmidt, K. M. (1999). A theory of fairness, competition, and cooperation. Quarterly Journal of Economics, 114(3), 817-868.

(22)

Fehr, E., & Fischbacher, U. (2003). The nature of human altruism. Nature, 425(6960), 785-791.

Giannotti, F., Pedreschi, D., Lukowicz, P., Kossmann, D., Crowley, J., & Helbing, D. (2013). A planetary nervous system for social mining and collective awareness. arXiv preprint arXiv: 1304.3700.

Hess, S., & Daly, A. (Eds.). (2014). Handbool< of Oioice Modelling. Edward Elgar Publishing.

Hensher, D. A., Rose, J., & Bertoia, T. (2007). The implications on willingness to pay of a stochastic treatment of attribute processing in stated choice studies. Transportation Research Part E: Logistics and Transportation

Review, 4J(2), 73-89.

Jakobsson, C, Fujii, S., & Garling, T. (2000). Determinants of private car users' acceptance of road pricing. Transport Policy, 7(2), 153-158.

Kelman, S. (1981). Cost-benefit analysis: an ethical critique. Regulation, 5, 33. Kivetz, R., Netzer, 0., & Srinivasan, V. (2004). Alternative models for capturing the compromise effect. Journal of Marketing Research, 41(3), 237-257. Leong, W., & Hensher, D. A. (2014). Contrasts of Relative Advantage

Maximisation with Random Utility Maximisation and Regret Minimisation.

Journal of Transport Economics and Policy (JTEP), 48(3), 167-186.

Manville, M., & Cummins, B. (in press). Why do voters support public

transportation! Public choices and private behavior. Transportation

McFadden, D. (2001). Economic choices. American Economic Review, 91(3), 351-378.

Nobelprize.org (2000). The Sveriges Riksbank Prize in Economic Sciences in

Memory of Alfred Nobel 2000 (James J. Heckman, Daniel L. McFadden)

- Popular information.

Nyborg, K. (2000). Homo economicus and homo politicus: Interpretation and aggregation of environmental values. Journal of Economic Behavior &

Organization, 42(3), 305-322.

Pet Shop Boys (1987). You are always on my mind, (album: Introspective) Simonson, I. (1989). Choice based on reasons: The case of attraction and

compromise effects. Journal of Consumer Research, 16, 158-174.

Statistical Software (2014). Upgrade manual for Latent GOLD Choice 5.0. Belmont, MA

Steenbruggen, J., Borzacchiello, M. T, Nijkamp, P., & Scholten, H. (2013). Data from telecommunication networks for incident management: An exploratory review on transport safety and security. Transport Policy, 28, 86-102. Train, K. E. (2009). Discrete choice methods with simulation. Cambridge

University Press.

Tversky, A., & Simonson, I. (1993). Context-dependent preferences.

(23)

Van Cranenburgh, S., Guevara, C.A., & Chorus, C.G. (in press). New insights on random regret minimization models. Transportation Researcli Part A Van Wee, B. (2011). Transport and ethics: Ethics and the evaluation of

transport policies and projects. Edward Elgar Publishing.

Wu, W. X., & Huang, H. J. (2010). A new model for studying the System Optimum-based pre-trip information release strategy and route choice behaviour. Transportmetrica, 6(4), 271-290.

(24)
(25)
(26)

Cytaty

Powiązane dokumenty

The model is capable of reproduc- ing the combined on/offshore events observed in the Duck94 and Hasaki data sets (Figures 5c and 2e, respec- tively), as well as of predicting

Further, tracking tracer fluxes showed that the various components of a model can be char- acterized by fundamentally different water age distributions which may be highly sensitive

55 Principatus: Grottcoviensis (ad Episcopum), Vratislaviensis, Schwi- dnicens s, Jauroviensis, Glogoviensis, Oppoliensis, Ratiboriensis, Ligni- cens's, Briegensis,

The choice of the optimal spherical radial basis function (SRBF) in local gravity field modelling from terrestrial gravity data is investigated.. Various types of SRBFs are

moments and quantiles of the empirical distribution, so they are estimators of the corresponding theoretical values. sample mean = estimator of the expected

- On the Existence of a Linear Connection so as a Given Tensor Field of the Type (1,1) is Parallel with Respect to This Connection O istnieniu koneksji liniowej takiej,

In the proof of this theorem, the key role is played by an effective interpretation of the well-known fact that an irreducible polynomial which is reducible over the algebraic

Augment a given network graph with an additional nullor (whose nullator branch and norator branch, respectively, connect the nodes κ 1 and κ 2 with the reference node), and then